The aim of this study is to employ machine learning (ML) in providing high-accuracy prediction of Cr(VI) removal efficiency by nickel hydroxide (n-Ni(OH)2) unconventional sorbent, towards the new era of artificial intelligence (AI) applications in (waste)water treatment. Hence, a reliable ML modeling was conducted based on the experimental investigation, considering different reaction parameters, including n-Ni(OH)2 dosage, initial pH, reaction temperature, and initial Cr(VI) concentration. Linear regression model was selected as the suitable regression model with respect to the obtained reasonable correlation and the less training time and evaluation time, comparing to other considered regression techniques. The adopted linear regression model, for the time-corresponding Cr(VI) removal efficiencies, exhibited satisfactory prediction accuracy. Furthermore, the importance of models’ coefficients was determined and implied the high importance of the dosage feature. In contrast, the initial concentration feature was significantly crucial at the early stage of the reaction (5–30 min) more than that at the late stage (60–120 min). The contributive effect (%) of the investigated features was mainly concentrated at the early stage of the reaction (5–10 min), with an average range of 50–80 %, which was in agreement with the experimental findings of the rapid and full removal of Cr(VI) by n-Ni(OH)2. The elucidated insights into the effects of different factors that influence Cr(VI) removal process by n-Ni(OH)2 revealed the underlying interactions and removal pathways, which shall benefit other researchers in the preliminary design of pilot-scale applications and anticipating the predicted performance.
All Science Journal Classification (ASJC) codes
- Analytical Chemistry
- Filtration and Separation